fast-uri 4.1.1 → 4.1.3

package/index.js

@@ -1,8 +1,23 @@
 'use strict'
 
-const { normalizeIPv6, removeDotSegments, recomposeAuthority, normalizePercentEncoding, normalizePathEncoding, normalizeQueryFragmentEncoding, escapePreservingEscapes, reescapeHostDelimiters, isIPv4, nonSimpleDomain } = require('./lib/utils')
+const { normalizeIPv6, removeDotSegments, recomposeAuthority, normalizePercentEncoding, normalizePathEncoding, serializePathEncoding, normalizeQueryFragmentEncoding, encodeQuery, encodeFragment, reescapeHostDelimiters, isIPv4, nonSimpleDomain } = require('./lib/utils')
 const { SCHEMES, getSchemeHandler } = require('./lib/schemes')
 
+const VALID_SCHEME = /^[A-Za-z][A-Za-z0-9+.-]*$/u
+const MALFORMED_SCHEME_ERROR = 'URI scheme is malformed.'
+
+/**
+ * @param {string} scheme
+ * @returns {string}
+ */
+function decodeValidScheme (scheme) {
+  const decodedScheme = unescape(String(scheme))
+  if (!VALID_SCHEME.test(decodedScheme)) {
+    throw new TypeError(MALFORMED_SCHEME_ERROR)
+  }
+  return decodedScheme
+}
+
 /**
  * @template {import('./types/index').URIComponent|string} T
  * @param {T} uri
@@ -26,7 +41,50 @@ function normalize (uri, options) {
  */
 function resolve (baseURI, relativeURI, options) {
   const schemelessOptions = options ? Object.assign({ scheme: 'null' }, options) : { scheme: 'null' }
-  const resolved = resolveComponent(parse(baseURI, schemelessOptions), parse(relativeURI, schemelessOptions), schemelessOptions, true)
+  const {
+    parsed: baseParsed,
+    malformedAuthorityOrPort: baseMalformed,
+    malformedPercentEncoding: baseMalformedPercentEncoding,
+    malformedSchemeSpecific: baseMalformedSchemeSpecific,
+    malformedHost: baseMalformedHost,
+    malformedScheme: baseMalformedScheme
+  } = parseWithStatus(baseURI, schemelessOptions)
+  const {
+    parsed: relativeParsed,
+    malformedAuthorityOrPort: relativeMalformed,
+    malformedPercentEncoding: relativeMalformedPercentEncoding,
+    malformedSchemeSpecific: relativeMalformedSchemeSpecific,
+    malformedHost: relativeMalformedHost,
+    malformedScheme: relativeMalformedScheme
+  } = parseWithStatus(relativeURI, schemelessOptions)
+  if (
+    baseMalformed ||
+    relativeMalformed ||
+    baseMalformedPercentEncoding ||
+    relativeMalformedPercentEncoding ||
+    baseMalformedSchemeSpecific ||
+    relativeMalformedSchemeSpecific ||
+    baseMalformedHost ||
+    relativeMalformedHost ||
+    baseMalformedScheme ||
+    relativeMalformedScheme
+  ) {
+    throw new Error(baseParsed.error || relativeParsed.error || 'URI is malformed.')
+  }
+  const resolved = resolveComponent(baseParsed, relativeParsed, schemelessOptions, true)
+  const resolvedSchemeHandler = getSchemeHandler((options && options.scheme) || resolved.scheme)
+  const resolvedHost = resolved.host
+  const resolvedHostIsIP = resolvedHost !== undefined && resolvedHost !== '' &&
+    (isIPv4(resolvedHost) || normalizeIPv6(resolvedHost).isIPV6)
+  canonicalizeHost(resolved, options || {}, resolvedSchemeHandler, resolvedHostIsIP)
+  // Percent escapes in an ASCII reg-name are encoded data. The WHATWG hostname
+  // parser can reject them even though fast-uri preserves them safely as RFC
+  // 3986 data. A raw non-ASCII host must still fail closed if conversion fails.
+  const encodedASCIIHost = resolvedHost && resolvedHost.indexOf('%') !== -1 &&
+    !/\P{ASCII}/u.test(resolvedHost)
+  if (resolved.error && !encodedASCIIHost) {
+    throw new Error(resolved.error)
+  }
   schemelessOptions.skipEscape = true
   return serialize(resolved, schemelessOptions)
 }
@@ -109,7 +167,7 @@ function equal (uriA, uriB, options) {
   const normalizedA = normalizeComparableURI(uriA, options)
   const normalizedB = normalizeComparableURI(uriB, options)
 
-  return normalizedA !== undefined && normalizedB !== undefined && normalizedA.toLowerCase() === normalizedB.toLowerCase()
+  return normalizedA !== undefined && normalizedB !== undefined && normalizedA === normalizedB
 }
 
 /**
@@ -132,30 +190,39 @@ function serialize (cmpts, opts) {
     reference: cmpts.reference,
     resourceName: cmpts.resourceName,
     secure: cmpts.secure,
+    to: cmpts.to,
+    subject: cmpts.subject,
+    body: cmpts.body,
+    headers: cmpts.headers,
     error: ''
   }
   const options = Object.assign({}, opts)
   const uriTokens = []
 
+  if (component.scheme) {
+    component.scheme = decodeValidScheme(component.scheme)
+  }
+
   // find scheme handler
   const schemeHandler = getSchemeHandler(options.scheme || component.scheme)
 
   // perform scheme specific serialization
   if (schemeHandler && schemeHandler.serialize) schemeHandler.serialize(component, options)
 
+  const hasAuthority = component.userinfo !== undefined || component.host !== undefined || component.port !== undefined
+  const pathNoScheme = !options.skipEscape && component.scheme === undefined && !hasAuthority
+
   if (component.path !== undefined) {
     if (!options.skipEscape) {
-      component.path = escapePreservingEscapes(component.path)
-
-      if (component.scheme !== undefined) {
-        component.path = component.path.split('%3A').join(':')
-      }
+      component.path = serializePathEncoding(component.path, pathNoScheme)
     } else {
       component.path = normalizePercentEncoding(component.path)
     }
   }
 
   if (options.reference !== 'suffix' && component.scheme) {
+    // Scheme handlers may replace the scheme during serialization.
+    component.scheme = decodeValidScheme(component.scheme)
     uriTokens.push(component.scheme, ':')
   }
 
@@ -178,6 +245,13 @@ function serialize (cmpts, opts) {
       s = removeDotSegments(s)
     }
 
+    // Dot-segment removal can expose a colon that was not originally in the
+    // first segment (for example, "./a:b"). Reapply path-noscheme encoding so
+    // the serialized relative reference cannot be reparsed as a URI scheme.
+    if (pathNoScheme) {
+      s = serializePathEncoding(s, true)
+    }
+
     if (
       authority === undefined &&
       s[0] === '/' &&
@@ -191,11 +265,11 @@ function serialize (cmpts, opts) {
   }
 
   if (component.query !== undefined) {
-    uriTokens.push('?', component.query)
+    uriTokens.push('?', encodeQuery(component.query))
   }
 
   if (component.fragment !== undefined) {
-    uriTokens.push('#', component.fragment)
+    uriTokens.push('#', encodeFragment(component.fragment))
   }
   return uriTokens.join('')
 }
@@ -206,6 +280,15 @@ const URI_PARSE = /^(?:([^#/:?]+):)?(?:\/\/((?:([^#/?@]*)@)?(\[[^#/?\]]+\]|[^#/:
 // with or without a scheme prefix, for the literal-backslash rejection below.
 const AUTHORITY_PREFIX = /^(?:[^#/:?]+:)?\/\/([^/?#]*)/
 
+// Captures the leading authority-introducer region after an optional scheme: a
+// run of forward slashes, backslashes, and the characters the WHATWG URL parser
+// removes before parsing (TAB U+0009, LF U+000A, CR U+000D). A valid introducer
+// is exactly "//". Node treats "\" as "/" on special schemes and strips those
+// characters first, so forms like "\\", "/\", "\/", "/<TAB>/", or a leading
+// "<TAB>//" reach an authority in Node while fast-uri's URI_PARSE folds them into
+// the path group (host confusion / SSRF / redirect bypass).
+const AUTHORITY_INTRODUCER_REGION = /^(?:[^#/:?]+:)?([/\\\t\n\r]*)/
+
 /**
  * @param {import('./types/index').URIComponent} parsed
  * @param {RegExpMatchArray} matches
@@ -223,10 +306,74 @@ function getParseError (parsed, matches) {
   return undefined
 }
 
+/**
+ * Checks percent syntax without decoding the represented octets. RFC 3986
+ * percent-encoding is byte-oriented, so sequences such as `%FF` are valid even
+ * though they are not independently valid UTF-8.
+ *
+ * @param {string|undefined} component
+ * @returns {boolean}
+ */
+function hasMalformedPercentEncoding (component) {
+  if (component === undefined) return false
+
+  let percent = component.indexOf('%')
+  while (percent !== -1) {
+    if (percent + 2 >= component.length || !/^[\da-f]{2}$/iu.test(component.slice(percent + 1, percent + 3))) {
+      return true
+    }
+    percent = component.indexOf('%', percent + 3)
+  }
+
+  return false
+}
+
+/**
+ * @param {RegExpMatchArray} matches
+ * @returns {boolean}
+ */
+function hasMalformedComponentPercentEncoding (matches) {
+  // Bracketed IP literals use a raw "%" as the zone separator for historical
+  // compatibility. Their parsing is intentionally left to normalizeIPv6.
+  const host = matches[4]
+  return hasMalformedPercentEncoding(matches[3]) ||
+    (host !== undefined && !(host[0] === '[' && host[host.length - 1] === ']') && hasMalformedPercentEncoding(host)) ||
+    hasMalformedPercentEncoding(matches[6]) ||
+    hasMalformedPercentEncoding(matches[7]) ||
+    hasMalformedPercentEncoding(matches[8])
+}
+
+/**
+ * @param {import('./types/index').URIComponent} parsed
+ * @param {import('./types/index').Options} options
+ * @param {{ domainHost?: boolean, unicodeSupport?: boolean }|undefined} schemeHandler
+ * @param {boolean} isIP
+ * @returns {boolean} whether host conversion failed
+ */
+function canonicalizeHost (parsed, options, schemeHandler, isIP) {
+  if (
+    !options.unicodeSupport &&
+    (!schemeHandler || !schemeHandler.unicodeSupport) &&
+    parsed.host &&
+    parsed.host[0] !== '[' &&
+    (options.domainHost || (schemeHandler && schemeHandler.domainHost)) &&
+    isIP === false &&
+    nonSimpleDomain(parsed.host)
+  ) {
+    try {
+      parsed.host = new URL('http://' + parsed.host).hostname
+    } catch (e) {
+      parsed.error = parsed.error || "Host's domain name can not be converted to ASCII: " + e
+      return true
+    }
+  }
+  return false
+}
+
 /**
  * @param {string} uri
  * @param {import('./types/index').Options} [opts]
- * @returns {{ parsed: import('./types/index').URIComponent, malformedAuthorityOrPort: boolean }}
+ * @returns {{ parsed: import('./types/index').URIComponent, malformedAuthorityOrPort: boolean, malformedPercentEncoding: boolean, malformedSchemeSpecific: boolean, malformedHost: boolean, malformedScheme: boolean }}
  */
 function parseWithStatus (uri, opts) {
   const options = Object.assign({}, opts)
@@ -242,6 +389,11 @@ function parseWithStatus (uri, opts) {
   }
 
   let malformedAuthorityOrPort = false
+  let malformedPercentEncoding = false
+  let malformedSchemeSpecific = false
+  let malformedHost = false
+  let malformedIPLiteral = false
+  let malformedScheme = false
 
   let isIP = false
   if (options.reference === 'suffix') {
@@ -265,11 +417,33 @@ function parseWithStatus (uri, opts) {
     malformedAuthorityOrPort = true
   }
 
+  // Reject a malformed or whitespace-smuggled authority introducer. fast-uri
+  // only recognizes a literal "//"; anything else in the leading separator run
+  // (a backslash, or a "//" that appears only after removing the TAB/LF/CR that
+  // Node strips) means the authority fast-uri parses differs from the one Node's
+  // URL resolves. Reject rather than rewrite, mirroring the literal-backslash
+  // guard above. Percent-encoded forms (%5C, %09) are untouched, valid data.
+  const introducerMatch = uri.match(AUTHORITY_INTRODUCER_REGION)
+  if (introducerMatch !== null) {
+    const region = introducerMatch[1]
+    const normalizedRegion = region.replace(/[\t\n\r]/g, '')
+    // Two or more leading separators introduce an authority.
+    if (normalizedRegion.length >= 2) {
+      if (normalizedRegion.slice(0, 2) !== '//') {
+        parsed.error = parsed.error || 'URI authority must not contain a literal backslash.'
+        malformedAuthorityOrPort = true
+      } else if (region.length !== normalizedRegion.length) {
+        parsed.error = parsed.error || 'URI authority introducer must not contain whitespace.'
+        malformedAuthorityOrPort = true
+      }
+    }
+  }
+
   const matches = uri.match(URI_PARSE)
 
   if (matches) {
     // store each component
-    parsed.scheme = matches[1] === undefined ? undefined : matches[1].toLowerCase()
+    parsed.scheme = matches[1]
     parsed.userinfo = matches[3]
     parsed.host = matches[4]
     parsed.port = parseInt(matches[5], 10)
@@ -277,6 +451,21 @@ function parseWithStatus (uri, opts) {
     parsed.query = matches[7]
     parsed.fragment = matches[8]
 
+    if (parsed.scheme !== undefined) {
+      const decodedScheme = unescape(parsed.scheme)
+      if (VALID_SCHEME.test(decodedScheme)) {
+        parsed.scheme = decodedScheme.toLowerCase()
+      } else {
+        parsed.error = parsed.error || MALFORMED_SCHEME_ERROR
+        malformedScheme = true
+      }
+    }
+
+    malformedPercentEncoding = hasMalformedComponentPercentEncoding(matches)
+    if (malformedPercentEncoding) {
+      parsed.error = parsed.error || 'URI contains malformed percent-encoding.'
+    }
+
     // fix port number
     if (isNaN(parsed.port)) {
       parsed.port = matches[5]
@@ -291,9 +480,16 @@ function parseWithStatus (uri, opts) {
     if (parsed.host) {
       const ipv4result = isIPv4(parsed.host)
       if (ipv4result === false) {
+        const bracketedIPLiteral = parsed.host[0] === '[' && parsed.host[parsed.host.length - 1] === ']'
         const ipv6result = normalizeIPv6(parsed.host)
-        parsed.host = ipv6result.host.toLowerCase()
-        isIP = ipv6result.isIPV6
+        isIP = ipv6result.isIPV6 || ipv6result.isIPVFuture === true
+        malformedIPLiteral = bracketedIPLiteral && ipv6result.error === true
+        parsed.host = isIP ? ipv6result.host : ipv6result.host.toLowerCase()
+
+        if (malformedIPLiteral) {
+          parsed.error = parsed.error || 'URI host is malformed.'
+          malformedAuthorityOrPort = true
+        }
       } else {
         isIP = true
       }
@@ -316,27 +512,14 @@ function parseWithStatus (uri, opts) {
     // find scheme handler
     const schemeHandler = getSchemeHandler(options.scheme || parsed.scheme)
 
-    // check if scheme can't handle IRIs
-    if (!options.unicodeSupport && (!schemeHandler || !schemeHandler.unicodeSupport)) {
-      // if host component is a domain name
-      if (parsed.host && (options.domainHost || (schemeHandler && schemeHandler.domainHost)) && isIP === false && nonSimpleDomain(parsed.host)) {
-        // convert Unicode IDN -> ASCII IDN
-        try {
-          parsed.host = new URL('http://' + parsed.host).hostname
-        } catch (e) {
-          parsed.error = parsed.error || "Host's domain name can not be converted to ASCII: " + e
-        }
-      }
-      // convert IRI -> URI
-    }
+    // convert Unicode IDN -> ASCII IDN when the effective scheme uses domain hosts
+    malformedHost = canonicalizeHost(parsed, options, schemeHandler, isIP)
 
     if (!schemeHandler || (schemeHandler && !schemeHandler.skipNormalize)) {
       if (uri.indexOf('%') !== -1) {
-        if (parsed.scheme !== undefined) {
-          parsed.scheme = unescape(parsed.scheme)
-        }
-        if (parsed.host !== undefined) {
-          parsed.host = reescapeHostDelimiters(unescape(parsed.host), isIP)
+        if (parsed.host !== undefined && !malformedIPLiteral) {
+          const host = isIP ? parsed.host : normalizePercentEncoding(parsed.host, true)
+          parsed.host = reescapeHostDelimiters(host, isIP)
         }
       }
       if (parsed.path) {
@@ -353,11 +536,14 @@ function parseWithStatus (uri, opts) {
     // perform scheme specific parsing
     if (schemeHandler && schemeHandler.parse) {
       schemeHandler.parse(parsed, options)
+      if (schemeHandler === SCHEMES.urn && parsed.nid === undefined) {
+        malformedSchemeSpecific = true
+      }
     }
   } else {
     parsed.error = parsed.error || 'URI can not be parsed.'
   }
-  return { parsed, malformedAuthorityOrPort }
+  return { parsed, malformedAuthorityOrPort, malformedPercentEncoding, malformedSchemeSpecific, malformedHost, malformedScheme }
 }
 
 /**
@@ -381,13 +567,17 @@ function normalizeString (uri, opts) {
 /**
  * @param {string} uri
  * @param {import('./types/index').Options} [opts]
- * @returns {{ normalized: string, malformedAuthorityOrPort: boolean }}
+ * @returns {{ normalized: string, malformedAuthorityOrPort: boolean, malformedPercentEncoding: boolean, malformedSchemeSpecific: boolean, malformedHost: boolean, malformedScheme: boolean }}
  */
 function normalizeStringWithStatus (uri, opts) {
-  const { parsed, malformedAuthorityOrPort } = parseWithStatus(uri, opts)
+  const { parsed, malformedAuthorityOrPort, malformedPercentEncoding, malformedSchemeSpecific, malformedHost, malformedScheme } = parseWithStatus(uri, opts)
   return {
-    normalized: malformedAuthorityOrPort ? uri : serialize(parsed, opts),
-    malformedAuthorityOrPort
+    normalized: malformedAuthorityOrPort || malformedPercentEncoding || malformedSchemeSpecific || malformedHost || malformedScheme ? uri : serialize(parsed, opts),
+    malformedAuthorityOrPort,
+    malformedPercentEncoding,
+    malformedSchemeSpecific,
+    malformedHost,
+    malformedScheme
   }
 }
 
@@ -397,14 +587,18 @@ function normalizeStringWithStatus (uri, opts) {
  * @returns {string|undefined}
  */
 function normalizeComparableURI (uri, opts) {
-  if (typeof uri === 'string') {
-    const { normalized, malformedAuthorityOrPort } = normalizeStringWithStatus(uri, opts)
-    return malformedAuthorityOrPort ? undefined : normalized
+  if (typeof uri !== 'string' && typeof uri !== 'object') {
+    return undefined
   }
 
-  if (typeof uri === 'object') {
-    return serialize(uri, opts)
+  let value
+  try {
+    value = typeof uri === 'string' ? uri : serialize(uri, opts)
+  } catch {
+    return undefined
   }
+  const { normalized, malformedAuthorityOrPort, malformedPercentEncoding, malformedSchemeSpecific, malformedHost, malformedScheme } = normalizeStringWithStatus(value, opts)
+  return malformedAuthorityOrPort || malformedPercentEncoding || malformedSchemeSpecific || malformedHost || malformedScheme ? undefined : normalized
 }
 
 const fastUri = {